train json structure

[0Crave]

Thank you for your job! I noticed that the training JSON structure you provided seems to be similar to the inference JSON structure. I have attempted to create my own training JSON based on your code, but the results have not been satisfactory. Could you please provide an example of the training JSON structure?

[cbstyle]

We uploaded the demo format of training data, benchmark/demo_grit20M_training_dataset.json. Refer to the field description of the original grit-20m dataset or the code in dataset/grit.py

[0Crave]

Thank you for your swift and helpful response.

[0Crave]

Thank you for your swift and helpful response. Your guidance on the training JSON structure was exactly what I needed to resolve my issue. I truly appreciate your support. Best regards, Bocheng Chen

…

------------------ 原始邮件 ------------------ 发件人: "Bo ***@***.***>; 发送时间: 2025年3月31日(星期一) 中午11:15 收件人: ***@***.***>; 抄送: ***@***.***>; ***@***.***>; 主题: Re: [360CVGroup/HiCo_T2I] train json structure (Issue #9) We uploaded the demo format of training data, benchmark/demo_grit20M_training_dataset.json. Refer to the field description of the original grit-20m dataset or the code in dataset/grid.py — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***> cbstyle left a comment (360CVGroup/HiCo_T2I#9) We uploaded the demo format of training data, benchmark/demo_grit20M_training_dataset.json. Refer to the field description of the original grit-20m dataset or the code in dataset/grid.py — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>